I did some research on network coding a few years ago (see [1] if interested), and I was very excited by the potential until I read the paper by Li, Li, and Lau that proves the maximum gain in throughput we can expect in undirected networks is 2x. https://iqua.ece.toronto.edu/papers/zp-transit.pdf This makes me think there is no point in introducing all this complexity... just stick to packet networks and routing, and you're at most 2x less efficient than then optimal netwrok coding scheme...

[1] A short literature review paper on network coding dating back to 2009: https://minireference.com/static/tmp/network_coding_review.p...